Knowledge
Optical Topography Measurement and Material Characterisation of Alu Cylinder Faces
ABSTRACT
Surface topography characterisation of engine cylinder walls using optical methods such as microscopic stripe projection, white light interferometry (WLI) and micro ellipsometry is outlined. Both the surface topography and also the kind of material can be measured, in especially the Si particle size and location distribution and the overall profile amplitude distribution.
In case of an iron coated aluminium matrix a simple method is described to quickly check the equal roughening of the surface and the pore size distribution. The measurement result decides whether rework is necessary before the time consuming and expensive spraying process is performed.
INTRODUCTION
In order to lower the friction between an aluminium cylinder wall and the piston ring either the wall is coated with a hard material by plasma spraying (this technique is used for smaller engines) or the aluminium matrix contains many small Si particles (AluSil or Silitec brand names) used in 8 -12 cylinder engines. Whereas in the first case it is vital for the adhesive property of the surface/coating interface that the surface is equally roughened before the spraying process and also no large pores are present it is important in the second case that the Si grains are distributed evenly and stick out of the surface by a constant amount resulting in a double peaked profile amplitude distribution.
For the quality control of cylinder surfaces only optical methods seem possible since the hard Si particles would quickly ruin the stylus of a conventional roughness tester and in the case of the spraying technique the roughened aluminium surface cannot be touched at all since the roughness aspherities are quite soft. Moreover, if an in-line inspection is required only optical methods can provide the necessary short cycle times.
GOALS OF OUR INSTRUMENT DEVELOPMENT
If a measurement method is to succeed in an industrial environment it must be easy to operate, lend itself to be used in-line and of course it has to deliver the required information about the surface. So far it seems difficult to achieve that optical methods conform to the roughness standards but in this application this isn’t necessary since surface features are to be checked for which no standards exist and moreover the advance of the technology is often so fast that there is no time to agree on standards.
The objectives to develop a new instrument were:
- ability to measure small surface features with lateral dimensions of a few micrometers
- ability of the instrument to find repeatedly the same measurement location on the cylinder wall
- ability to distinguish between different kinds of materials and correct for phase changes
Naturally the instrument should be easy to operate and able to check cylinders with a diameter starting from 65 mm.
In a careful decision process lasting several months several measurement methods were investigated in order to find out which one would lend itself best to fulfill the requirements as listed above and also to give the best fidelity of surface profile reproduction. It could be shown that excellent results were obtained with the WLI method but also the microscopic stripe projection is a good method and it is especially suited to be designed such, that in combination with an ellipsometric procedure also the material can be recognised in order to get both a high resolution profile and material mapping of the surface.
OPTICAL METHODS AND INSTRUMENTATION
During the past years an amazing variety of optical surface profile measurement methods evolved but all of them can easily be categorised due to the Heisenberg uncertainty principle. Fig. 1 in the appendix summarises the given situation and according to Fig. 2 a high resolution in the desired Z direction requires either a high numerical aperture of the sensor objective lens or using broadband light in order to fulfill the requirement for a large momentum which in turn allows a high definition position determination of the photon. In especially WLI allows more easily to measure surface features with a high aspect ratio and seems to produce a higher profile fidelity at steep slopes than other methods. Fig. 3 shows the working principle of WLI.
Fig. 1
Fig. 2
White Light Interferometer Principle
Fig. 3
We also tested a novel microscopic stripe projection technique which gave excellent results and which has been designed such that simultaneously a micro-ellipsometer was incorporated. The latter allows to determine the ellipsometric angles ? and ? which are specific for each material. Using these two methods both the surface profile and material distribution of the surface could be measured. This is of a big advantage when measuring silicon doped surfaces like AluSil or Silitec in order to make sure whether a surface aspherity really corresponds to a Si particle. As one can see from Fig. 4 it is quite common with optical profiling methods that obvious surface elevations are erroneously recorded as valleys in case the surface is made up of several different materials. Fig. 5 shows schematically the stripe projection method and Fig. 6 the actual instrument. When not in use it rests on a stand which also provides standards to check the calibration of the measured profile amplitudes and the field of view.
Oil droplets on gauge block.
Fig. 4
Fig. 5
White Light Interferometer surface profiler for cylinders.
Fig. 6
A unique advantage of the instrument is that it features two microscope objectives. The measurement objective can easily be changed in order to accommodate different magnifications and field of views. The second objective is fixed and provides a large field of view of approx. 4 mm² in order to get a good overview of the location to be measured. Another important feature are 3 motorised axes to move the measurement spot to any desired position on the cylinder wall. This movement is highly accurate due to the glass scales to measure the actual position in Y, R, Z.
RESULTS
A customer will use this instrument at first in the development department and after establishing the findings about the surface characteristics this method will serve for quality control purposes in-line. The experiences made so far show that it is no real alternative to cut an engine in order to be able to use a standard lab profilometer. Instead the main advantage of this instrument results from it’s ability to measure and inspect surface features of interest, e.g. pores and Si grains repeatedly after running the engine and disassembly. Thereby not only the usual wear can be observed but also valuable information be collected under what circumstances Si grains brake out of the surface leaving unwanted pits and what kind of pores will enlarge with time while running the engine. Unexpected and very valuable findings were gained but under standably we are not allowed to report about these in detail. Fig. 11a and Fig. 11b give an example for a surface measurement before and after running the engine. Clearly one sees that one can find the same surface area although the instrument was removed in between the measurements. Fig. 11a shows the familiar honing structure and Fig. 11b the same surface section after running the engine for many hours. The surface structure seems to be smoothened and vertical scratches have been generated. The profile images shown here demonstrate the potential of this new instrument to help the engineer to understand and improve the surface characteristics in order to minimise the friction and ensure good lubrication.
Area profile of a new cylinder showing the honig structure.
Fig. 11a
The same area as in Fig. 11a but after running the engine.
Fig. 11b
Fig. 12
Grey cast iron, polished.
Fig. 13
The silicon grains should be distributed equally across the surface and also they should stick out by a similar amount. If single grains stick out more than the average they will be shaved off by the piston ring.
Fig. 7 +
Fig. 8 show typical area surface profile images. The instrument features objectives with magnifications of 50x, 20x, 10x.
The low magnification objective provides a larger measurement area but since the numerical aperture is smaller it’s ability to collect light is limited and therefore one will see more dark areas where no data could be collected. But some workpieces still deliver good results as is demonstrated in
Fig. 9 +
Fig. 10.
Therefore it makes more sense to stitch basic measurement fields together while taking advantage of the good resolution and light collecting ability of the high numerical objective. Naturally it takes some time to make all the measurements but the instrument provides all necessary hard- and software features for stitching also curved images of a cylinder.
False colour area profile with Si particles and holes where particles have been torn out.
Fig. 7
Area profile measurement with honing marks of the aluminium matrix.
Fig. 8
Stitched image of individual measurement field sizes of 250 x 250 µm.
Fig. 9
False colour area profile showing Si grain cluster formation.
Fig. 10
When calculating the amplitude distribution from the profile data one will see a curve showing two peaks whereby one corresponds to the metal matrix surface and the other to the Si particle plateau. From the distance between the peaks the average grain height can be deducted.
MATERIAL CORRECTION
In case the surface under measurement is made up of more than one material all known optical profilers give more or less wrong results (see reference for mathe matical details). These manifest themselves as artificial spikes at surface discontinuities and also as wrong surface amplitudes. The reason is that the equation de scribing the physics of light reflection from a surface contains several components, whereby some of them aren’t straightforward to be taken into account. In order to do so it is necessary to measure parameters like the polarisation state of the ingoing and reflected light beam and this is difficult to achieve in a pixelwise manner in ordinary surface profiling instruments. Also the common ellipsometers aren’t designed such that the polarisation state of fractions of the probing light beam is determined. To achieve this we made a completely new design whereby a microscopic stripe projection instrument is combined with a microellipsometer delivering all measur able quantities which enter the complete equation governing the reflection of a light beam from a surface. By switching on and off the ellipsometer part one can easily see the error which an ordinary instrument has got when the profile of a non-uniform surface consisting of more than one material is being measured.
Different colours are clearly visible whereby each colour is a measure for the amount of a specific material at the corresponding location. One sees the graphite sections but also the rest of the surface isn’t that uniform with respect to the material as one would assume.
Fig. 14 shows similar conditions for a SiAl surface. Depending whether the silicon grains are exposed by etching or a mechanical grinding process more or less aluminium is smeared over the silicon grains and forms layers of different oxidation states.
Fig. 15 gives more insight in the measured profile amplitudes. The top image shows the profiles as measured by a profilometer but the bottom image shows that there are layers which give wrong profile amplitude results if not taken into account properly.
Fig. 16 shows more explicitly the data correction necessary to arrive at true profile amplitude values due to the modification of the light properties when being reflected by different materials.
Complex refractive index
Pixelwise measured complex refractive index of a typical AluSil surface. One can see the Si particles. But also it can be seen that the area corresponding to their surface isn’t coloured uniformly which means that the Si grains are partially covered with aluminium which may have been smeared over them during the honing process. But also the aluminium surface is not homogeneous but covered with some kind of coating, probably oxide.
Fig. 14
Cylinder bore, Si-crystals in Al
Thickness of the Overlayers [nm]
FIg. 15
The middle line shows the measured profile of a surface elevation consisting of a different material than the base line. The top line represents the corrected height profile. Correction is made with respect to the modification of the property of the light beam which it experiences when it is reflected by different materials.
Fig. 16
FAST SURFACE SCANNER
In contrast to the above task where small surface areas have to be inspected and profiles be measured with high accuracy there is another inspection task concerning smaller aluminium engines where the cylinder surface is coated by a spraying technique. The coating sticks to the metal surface simply by clamping to the surface irregularities. To enable the best possible clamping efficiency the surface must contain a lot of cavities and steep slopes which can be created by sand blasting or a water jet. After this treatment the surface looks grey and dull due to the extremely rough surface with amplitudes of approx. 80 µm. Large pores or too high a pore density or pore clusters in certain regions as well as no clean up in some areas will cause a deterioration of the clamping effect and be harmful for the proper function of the surface. In this state the surface cannot be touched since the delicate surface asperities would be damaged.
To ensure high product quality and at the same time avoid rejects it is requested to optically check the roughened surface for pores and no clean up areas. A similarly important requirement is that this check is done within a few seconds corresponding to the manufacturing cycle time. Also a high resolution is required, say about 50 µm for pore sizes. And at last the probing optical head should be able to operate also in small cylinders of about 60 mm or less.
This task cannot be solved by a standard camera setup. Not only that there would be a space problem for the camera plus imaging optics but also the high spatial resolution required in order to find small pores and no clean up areas could only be handled by a camera based instrument if a stitching process would be used which is impossible due to the time requirement. Therefore we developed an instrument using a scanning bar like in an office scanner (
Fig. 17). The in-line version features as many scanners as there are cylinders so the time to check a complete engine block is the same as for one cylinder only. The linear photo diode array covers the complete length of the cylinder and is guided along the wall by a suitable mechanical setup. The detector samples the surface with a resolution of 400-600 dpi and a customised image processing software determines the pore size distribution and no clean up areas. During an extensive test period it could be shown that the no clean up areas and the pores as detected by this instrument are directly linked to the adhesive strength of the sprayed coating. So the output of the surface inspection by this instrument is an accept/rework decision based on the detected number of pores, pore sizes, pore cluster formation, minimum distance between adjacent pores and size of no clean up areas. At the same time this instrument can equally be used to measure the honing angle and clean crossover of honing grooves or burrs congesting the honing groove. A modified version lends itself to check the deck face for scratches and pores.
Surface scanner and images of scanned cylinder surfaces.
Fig. 17
CONCLUSION
Two different kinds of instruments were discussed in this paper. Both describe solutions for different problems associated with the manufacture of modern aluminium engines. Ways to lower the friction is either to incorporate silicon particles into the metal matrix and have them stand proud of the surface by approx. 1-2 µm or put a hard coating onto the surface. The resulting surface structures are quite different and require different probing methods to check their quality and functionality.
A white light interferometer or a microscopic stripe projection instrument with an incorporated micro ellipsometer can be used to check small areas of AluSil or Silitec surfaces and measure the particle size and spacial distribution and breakage of the grains. The achievable accuracy depends on both the objective used and the surface structure. One has to distinguish between the lateral resolution which cannot be more than approx. 1 µm by the laws of optics and the vertical resolution which theoretically approaches 1 nm. Also there is a trade off between the desirable high numerical aperture of the objective and the desire to measure into deep grooves. In contrast to all other known optical displacement measurement methods the vertical resolution of a white light interferometer is not linked to the numerical aperture and therefore the magnification. But unfortunately instruments based on this method often suffer from poor illumination conditions since the available light sources aren’t ideal.
In the second part a novel surface scanner is described which lends itself to check the cylinder surface before the spraying process to ensure the absence of harmful pores or no clean up areas. In contrast to conventional camera based systems it can check the entire surface in one go and within seconds delivering at the same time a spacial resolution better than 100 µm.
ACKNOWLEDGEMENTS
We thank DaimlerChrysler and Federal Mogul for their support of this work.
